An integrated machine learning model for automatic road crack detection and classification in urban areas

Cracks in the asphalt are the first and most common deterioration type of roads that generally threaten the safety of roads and highways. In recent years, automated inspection has been considered due to the high cost and error of manual methods. For this purpose, different machine learning technique...

Full description

Saved in:
Bibliographic Details
Published in:The international journal of pavement engineering 2022-08, Vol.23 (10), p.3536-3552
Main Authors: Ahmadi, Abbas, Khalesi, Sadjad, Golroo, Amir
Format: Article
Language:English
Subjects:
Algorithms
Classification
Crack detection
Cracks
Decision trees
Feature extraction
Flaw detection
Heuristic methods
Hough transform
Hough transformation
Hybrid model
Image classification
Image processing
Image segmentation
Inspection
Machine learning
Neural networks
Roads
Urban areas
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Cracks in the asphalt are the first and most common deterioration type of roads that generally threaten the safety of roads and highways. In recent years, automated inspection has been considered due to the high cost and error of manual methods. For this purpose, different machine learning techniques have been developed. In this study, an integrated model is proposed, which involves the following steps: image segmentation, noise reduction, feature extraction, and crack classification. In the first two steps, heuristic algorithms are proposed, and then in the third step, the Hough transform technique and the heuristic equations are used to extract the main features of cracks. In the fourth step, six different classification models, including neural network, SVM, decision tree, KNN, Bagged Trees, and a proposed hybrid model, are implemented. Experimental results show that the proposed hybrid model can achieve more accurate results with 93.86% overall accuracy.
ISSN:1029-8436
1477-268X
DOI:10.1080/10298436.2021.1905808